Archive
新模型、開放權重、多模態、語音、影像與模型架構。
最新模型
Inception launched Mercury 3 on October 19, expanding its context window to one million tokens and claiming output speeds of up to 1,068 tokens per second. Application developers have begun integrating it, but the company’s speed and quality comparisons still await independent evaluation.
GitHub Repo
Hugging Face released Transformers 5.20.2 on October 19, fixing a bug that could discard input container types when moving inputs during generation and affect multimodal models. The release also formally includes a BF16 gradient patch for a specific expert-parallel configuration, providing a deployable version for an issue previously reported by the community.
GitHub Repo
A Transformers community report on October 19 says leftover position information may alter sparse attention selection when Qwen4-Exp reuses a reset cache. The report includes a reproduction and a local patch, but maintainers have not confirmed it, and evaluation with pretrained models remains incomplete.
GitHub Repo
Ollama released version 0.40.7-rc0 on October 18, adding MLX backend support for PPLX Decider v1.2. The update covers non-causal attention, decision readout, and image processing, but local performance and post-quantization probabilities still need validation.
GitHub Repo
An October 18 report from the Open WebUI community says the same data may enter model context twice when FastMCP tools return lists—as text and structured content. A fix was proposed that day but has not been merged; actual token and cost effects still need to be measured against workloads.
最新模型
On October 16, Kyutai announced REKI and released its pretrained and fine-tuned weights, along with data processing, training, and deployment code. The model adds short-term memory to its action expert; the team’s tests show fewer repetitive motions, but the weights are limited to non-commercial use and the results await independent replication.
GitHub Repo
Hugging Face released Transformers 5.20.1 on October 16, fixing a regression in 5.20.0 that affected model loading with PyTorch 2.7. The update raises the version threshold for distributed checkpoint features to PyTorch 2.8, preventing a missing-module error from disrupting ordinary loading workflows.
GitHub Repo
A Transformers community report on October 17 said some Mask2Former projection layers are stored in a plain Python list and may be omitted from optimizer and device-transfer workflows. A patch has been proposed, but the official fix release and any effect on segmentation accuracy remain unconfirmed.
最新模型
On October 15, IFM released six Jais 3 models ranging from 0.9B to 375B parameters, focused on Arabic reasoning and tool use. The release includes training-stage checkpoints and evaluation assets, though the performance results still await independent replication.
GitHub Repo
Ollama released version 0.40.6 on October 16, fixing an issue where inference subprocesses could continue running after its Windows command-line service was forcibly terminated. The update uses operating-system process lifecycle management and covers llama-server and the MLX runner.
最新模型
On October 16, NVIDIA released Dory 1.0, which runs shared layers repeatedly so the same 4B weights can support different inference compute budgets. The model offers a final-loop KV cache option, but performance results are still based on NVIDIA’s own tests, and integration with major inference frameworks remains pending.
最新模型
Elorian launched the Glimpse research preview on October 16, enabling the model to zoom in on images, revisit video, and use multimodal tools during reasoning. The model is available on API platforms, but official evaluations use different agents and reasoning settings; real-world latency and cost still need validation.